Is Llama 4 Scout better than Ministral 3 8B 2512?

On accuracy they are statistically tied: Llama 4 Scout scored 41.2% and Ministral 3 8B 2512 42.5% on the same 80 questions from open datasets. Similar accuracy; Ministral 3 8B 2512 costs less per question.

Which is cheaper, Llama 4 Scout or Ministral 3 8B 2512?

Ministral 3 8B 2512 is cheaper: $0.00027 per question as billed, against $0.00037. List prices: Llama 4 Scout $0.1 in / $0.3 out, Ministral 3 8B 2512 $0.15 in / $0.15 out per million tokens.

Which is faster, Llama 4 Scout or Ministral 3 8B 2512?

Ministral 3 8B 2512 is faster at 124 output tokens per second, against 15. Time to first token: 1.2 s for Llama 4 Scout, 0.3 s for Ministral 3 8B 2512.

Which has the larger context window, Llama 4 Scout or Ministral 3 8B 2512?

Llama 4 Scout, with 1.3M tokens against 262k.

Head to head · tested 27 Oct 2026

Llama 4 Scout vs Ministral 3 8B 2512: accuracy, speed and price

Similar accuracy; Ministral 3 8B 2512 costs less per question.

MetaLlama 4 Scout41.2% score · #54$0.00037 per question15 tok/s

MistralMinistral 3 8B 251242.5% score · #52$0.00027 per question124 tok/s

Verdict

Built from the measurements below

Accuracy is too close to call: Llama 4 Scout scored 41.2% and Ministral 3 8B 2512 42.5% on the same 80 questions, a gap within what chance alone could produce.
Ministral 3 8B 2512 is 1.4× cheaper per question as billed ($0.00027 vs $0.00037), reasoning tokens included.
Ministral 3 8B 2512 generates text 8.1× faster: 124 tokens per second against 15.
Ministral 3 8B 2512 starts answering sooner: 0.3 s to the first token, against 1.2 s for Llama 4 Scout.
Llama 4 Scout accepts more context: 1.3M tokens against 262k.

Which should you choose?

Choose Llama 4 Scout for shorter outputs (about 723 tokens per question against 1,496) and a 1.3M-token context window. Choose Ministral 3 8B 2512 for the lower cost per correct answer ($0.00063 against $0.0009), faster streaming and a quicker first token. The widest subject gap is in economics: Ministral 3 8B 2512 answered 7 of 9 correctly, Llama 4 Scout 4 of 9. With so few questions per subject, treat this as a lead to test on your own prompts, not a result.

Accuracy gap -1.3 points (z = -0.16). A winner is called only when z is at least 1.96, the usual 95% threshold. Speed, latency and price need a 10% margin.

Head to head

Same questions, same prompt, same API

Measure	Llama 4 Scout	Ministral 3 8B 2512	Edge
Verification scoreHigher is better	41.2%	42.5%	Too close to call
Cost per questionLower is better	$0.00037	$0.00027	Ministral 3 8B 2512
Output speedHigher is better	15	124	Ministral 3 8B 2512
Time to first tokenLower is better	1.2 s	0.3 s	Ministral 3 8B 2512
End-to-end response timeLower is better	25.6 s	3.5 s	Ministral 3 8B 2512
Tokens per questionLower is better	723	1,496	Llama 4 Scout
Context windowHigher is better	1.3M	262k	Llama 4 Scout

Side by side

Bars are scaled to the larger of the two values. The winner on each measure is marked in bold.

Every measure

Llama 4 Scout first, Ministral 3 8B 2512 second

Verification scorehigher is better

Llama 4 Scout41.2%

Ministral 3 8B 251242.5%

Cost per questionlower is better

Llama 4 Scout$0.00037

Ministral 3 8B 2512$0.00027

Output speedhigher is better

Llama 4 Scout15

Ministral 3 8B 2512124

Time to first tokenlower is better

Llama 4 Scout1.2 s

Ministral 3 8B 25120.3 s

End-to-end response timelower is better

Llama 4 Scout25.6 s

Ministral 3 8B 25123.5 s

Tokens per questionlower is better

Llama 4 Scout723

Ministral 3 8B 25121,496

Context windowhigher is better

Llama 4 Scout1.3M

Ministral 3 8B 2512262k

Score vs. cost per question

Both models in color, every other model in grey

Results by subject and dataset

Questions answered correctly. Small samples: a gap of one or two questions is noise.

Subject	Questions	Llama 4 Scout	Ministral 3 8B 2512
Science	30	13/30	10/30
Engineering	15	2/15	4/15
Medicine	11	6/11	5/11
Economics	9	4/9	7/9
Law	6	2/6	1/6
Humanities & social science	9	6/9	7/9
SuperGPQA dataset	50	13/50	10/50
MMLU-Pro dataset	30	20/30	24/30

Specifications and pricing

From OpenRouter's catalogue, plus the leaderboard's test dates

Specification	Llama 4 Scout	Ministral 3 8B 2512
Creator	Meta	Mistral
Released	5 Apr 2025	2 Dec 2025
Weights	Open weights	Open weights
Reasoning	No	No
Input	text, image	text, image
Output	text	text
Context window	1.3M tokens	262k tokens
Knowledge cutoff	31 Aug 2024	—
List price in / out per 1M	$0.1 / $0.3	$0.15 / $0.15
Blended price per 1M (3:1)	$0.15	$0.15
Cheapest host	DeepInfra · $0.08 / $0.24	Mistral · $0.15 / $0.15
Hosts on OpenRouter	2	1
Last tested	27 Oct 2026	26 Oct 2026

Compare any two models →

Llama 4 Scout vs Ministral 3 8B 2512: frequently asked questions

Is Llama 4 Scout better than Ministral 3 8B 2512?: On accuracy they are statistically tied: Llama 4 Scout scored 41.2% and Ministral 3 8B 2512 42.5% on the same 80 questions from open datasets. Similar accuracy; Ministral 3 8B 2512 costs less per question.
Which is cheaper, Llama 4 Scout or Ministral 3 8B 2512?: Ministral 3 8B 2512 is cheaper: $0.00027 per question as billed, against $0.00037. List prices: Llama 4 Scout $0.1 in / $0.3 out, Ministral 3 8B 2512 $0.15 in / $0.15 out per million tokens.
Which is faster, Llama 4 Scout or Ministral 3 8B 2512?: Ministral 3 8B 2512 is faster at 124 output tokens per second, against 15. Time to first token: 1.2 s for Llama 4 Scout, 0.3 s for Ministral 3 8B 2512.
Which has the larger context window, Llama 4 Scout or Ministral 3 8B 2512?: Llama 4 Scout, with 1.3M tokens against 262k.